The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes user-defined 'tracks' which are interpolated into sub-agent prompts. This creates a potential surface for indirect prompt injection where instructions embedded in the track description could attempt to influence sub-agent behavior.
Ingestion points: tracks input array in SKILL.md.
Boundary markers: The skill uses explicit file manifests to limit sub-agent scope, though it does not explicitly use delimiters to wrap the track content in the sub-agent prompts.
Capability inventory: The orchestrator uses file_system tools and executes verification commands like go build, go test, and git status across the workspace.
Sanitization: No specific sanitization or filtering of the track description strings is performed before interpolation.
[COMMAND_EXECUTION]: The skill utilizes shell commands (git status, grep, go build, go test) to perform independent verification of agent output. These commands are part of the core functionality for a development-focused orchestration skill and are used to ensure code quality and prevent merge conflicts.

parallel-dispatch